Portrait of Amir Ozhan Dehghani

Amir Ozhan Dehghani

Montréal, QC

I’m an MSc student at Mila and McGill University.

Explore my work

Each glowing node in the brain is a project, grouped into three clusters: research, applied projects and community. Pick one to see what I did and what came of it.

News

Papers, roles and milestones, newest first.

  1. Excited to have my first public talk at Neurolingo.

  2. Started as an Applied AI Research Intern at Mila, working on the Canada Mineral Mapping initiative.

  3. Received the UNIQUE Excellence Scholarship.

  4. Ranked first for the Visual Research Science Award.

  5. Presented CB-SOM at ICLR 2025, supported by a UNIQUE Travel Award. See the project page

  6. Started my MSc in Computational Neuroscience at Mila and McGill.

Experience

  1. 2026

    Applied AI Research Intern

    Mila – Quebec AI Institute, Montréal

    Designed and deployed multimodal models for the Canada Mineral Mapping initiative.

  2. 2024–25

    Data Scientist

    AIVALON, Montréal

    Built coronary artery disease detection from HR, EKG and BVP signals.

  3. 2023–24

    Research Assistant

    Université de Montréal

    Led an end-to-end project testing neural network models of visual learning against human experiments built in PsychoJS.

  4. 2020–23

    Undergraduate Researcher

    McGill University

    Developed topographic deep networks and compared them with neurphysiological data.

Leadership

  • 2021–25

    PharmaHacks, President, VP Machine Learning and VP Sponsorship. Grew the team from 15+ memebers, ran the 2022,2023,2025 hackathons at Mila with 200+ participants, and secured $18K+ in sponsorship.

Education

  • 2025–27

    MSc, Computational Neuroscience, Mila and McGill University

  • 2019–23

    BSc Honours, Neuroscience, McGill University.

Publications

  1. 2026

    Local lateral connectivity is sufficient for replicating cortex-like topographical organization in deep neural networks

    Qian X., Dehghani A., Farahani A., Bashivan P. Nature Communications.

  2. 2025

    Credit-based self-organizing maps: training deep topographic networks with minimal performance degradation

    Dehghani A., Qian X., Farahani A., Bashivan P. ICLR 2025.

    Paper Project page Code OpenReview
  3. 2024

    Geometry of representations in artificial neural networks predicts visual learning specificity

    Dehghani A., Bakhtiari S. Healthy Brains, Healthy Lives Symposium.

    Abstract
  4. 2022

    Learning robust kernel ensembles with kernel average pooling

    Bashivan P., Ibrahim A., Dehghani A., Ren Y. arXiv.

    arXiv

Awards

  • 2026

    UNIQUE Excellence Scholarship

  • 2025

    Visual Research Science Award, ranked first

  • 2025

    Internal Graduate Studentship, MSc

  • 2025

    UNIQUE Travel Award

  • 2024

    Graduate Research Excellence Award

  • 2024

    Max E. and Jane K. Childress Entrance Fellowship in Physiology